jonatasgrosman/exp_w2v2r_de_vp-100k_age_teens-10_sixties-0_s362

.gitattributes

@@ -32,3 +32,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+cv7_transcriptions.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,17 @@
+---
+language:
+- de
+license: apache-2.0
+tags:
+- automatic-speech-recognition
+- de
+datasets:
+- mozilla-foundation/common_voice_7_0
+---
+# exp_w2v2r_de_vp-100k_age_teens-10_sixties-0_s362
+
+Fine-tuned [facebook/wav2vec2-large-100k-voxpopuli](https://huggingface.co/facebook/wav2vec2-large-100k-voxpopuli) for speech recognition using the train split of [Common Voice 7.0 (de)](https://huggingface.co/datasets/mozilla-foundation/common_voice_7_0).
+When using this model, make sure that your speech input is sampled at 16kHz.
+
+This model has been fine-tuned by the [HuggingSound](https://github.com/jonatasgrosman/huggingsound) tool.
+    

config.json

@@ -0,0 +1,116 @@
+{
+  "_name_or_path": "facebook/wav2vec2-large-100k-voxpopuli",
+  "activation_dropout": 0.05,
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": false,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.05,
+  "bos_token_id": 1,
+  "classifier_proj_size": 256,
+  "codevector_dim": 768,
+  "contrastive_logits_temperature": 0.1,
+  "conv_bias": true,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": false,
+  "diversity_loss_weight": 0.1,
+  "do_stable_layer_norm": true,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.05,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.05,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.05,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.05,
+  "mask_channel_length": 10,
+  "mask_channel_min_space": 1,
+  "mask_channel_other": 0.0,
+  "mask_channel_prob": 0.0,
+  "mask_channel_selection": "static",
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.05,
+  "mask_time_selection": "static",
+  "model_type": "wav2vec2",
+  "num_adapter_layers": 3,
+  "num_attention_heads": 16,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 24,
+  "num_negatives": 100,
+  "output_hidden_size": 1024,
+  "pad_token_id": 0,
+  "proj_codevector_dim": 768,
+  "tdnn_dilation": [
+    1,
+    2,
+    3,
+    1,
+    1
+  ],
+  "tdnn_dim": [
+    512,
+    512,
+    512,
+    512,
+    1500
+  ],
+  "tdnn_kernel": [
+    5,
+    3,
+    3,
+    1,
+    1
+  ],
+  "torch_dtype": "float32",
+  "transformers_version": "4.18.0",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 38,
+  "xvector_output_dim": 512
+}

cv7_evaluation.json

@@ -0,0 +1,396 @@
+{
+  "validation": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.06576375492714175,
+      "wer": 0.25034658040665436
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.06474856479065304,
+      "wer": 0.24779578889527187
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.061615654791771196,
+      "wer": 0.23355263157894737
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.07855343115458722,
+      "wer": 0.2812336644014637
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.06236189540878959,
+      "wer": 0.2619047619047619
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.07290897454831881,
+      "wer": 0.2737693863789616
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.06778632132739781,
+      "wer": 0.2609731627790319
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.0636215334420881,
+      "wer": 0.24390243902439024
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.06433513101269271,
+      "wer": 0.25604763676963155
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.09352548036758564,
+      "wer": 0.30523338048090526
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.0647323092759207,
+      "wer": 0.24773124635750562
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.07061680056951083,
+      "wer": 0.26329278887923546
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.10177404295051354,
+      "wer": 0.3597560975609756
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.06785848701543094,
+      "wer": 0.2498703991705547
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.0653891375167882,
+      "wer": 0.2495600536774691
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.09455958549222798,
+      "wer": 0.3504273504273504
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.030405405405405407,
+      "wer": 0.16666666666666666
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.06829268292682927,
+      "wer": 0.21212121212121213
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.16953316953316952,
+      "wer": 0.4576271186440678
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.06561168982404646,
+      "wer": 0.24780749835562377
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.06684846698113207,
+      "wer": 0.24619289340101522
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.03783783783783784,
+      "wer": 0.21153846153846154
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.10411622276029056,
+      "wer": 0.33884297520661155
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.10324232081911262,
+      "wer": 0.3422459893048128
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.059075342465753425,
+      "wer": 0.24848484848484848
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.09221556886227544,
+      "wer": 0.312
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.06388888888888888,
+      "wer": 0.2982456140350877
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.09931686810299527,
+      "wer": 0.3710247349823322
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.08080808080808081,
+      "wer": 0.30434782608695654
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.07948589967556775,
+      "wer": 0.2970639032815199
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.1213235294117647,
+      "wer": 0.425
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.07594937975290164,
+      "wer": 0.2736751009643122
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.07518392768708423,
+      "wer": 0.27158241674012185
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.07084493540877346,
+      "wer": 0.2594282238442822
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.12368870456208832,
+      "wer": 0.37225042301184436
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.11659541391371939,
+      "wer": 0.37193460490463215
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.06762280387235568,
+      "wer": 0.26104218362282877
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.07274867194706874,
+      "wer": 0.2754374594944913
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.13394495412844037,
+      "wer": 0.4931506849315068
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.08775794962179628,
+      "wer": 0.3066132264529058
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.10996171703184818,
+      "wer": 0.3429362880886427
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.0752898465079043,
+      "wer": 0.27173330064207674
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.0786661016550133,
+      "wer": 0.2814347339830644
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.1346153846153846,
+      "wer": 0.32954545454545453
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09353998910213719,
+      "wer": 0.33500837520938026
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.07603317119353034,
+      "wer": 0.2736121453429175
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.13479623824451412,
+      "wer": 0.47058823529411764
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.2053872053872054,
+      "wer": 0.5777777777777777
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bulgaria",
+      "cer": 0.0,
+      "wer": 0.0
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.25474254742547425,
+      "wer": 0.7547169811320755
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.03260869565217391,
+      "wer": 0.14285714285714285
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.04040404040404041,
+      "wer": 0.2413793103448276
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.07255139056831923,
+      "wer": 0.26141078838174275
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.09742236655165415,
+      "wer": 0.34173669467787116
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.13978494623655913,
+      "wer": 0.4666666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.15551839464882944,
+      "wer": 0.4673913043478261
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.08131394722670975,
+      "wer": 0.2871652816251154
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.08121827411167512,
+      "wer": 0.47619047619047616
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.05985915492957746,
+      "wer": 0.2894736842105263
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hungary",
+      "cer": 0.13333333333333333,
+      "wer": 0.5357142857142857
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.10475266731328807,
+      "wer": 0.35333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.07020006090537159,
+      "wer": 0.2593293672255273
+    },
+    {
+      "subset_type": "accent",
+      "subset": "denmark",
+      "cer": 0.058823529411764705,
+      "wer": 0.2222222222222222
+    },
+    {
+      "subset_type": "accent",
+      "subset": "greece",
+      "cer": 0.11538461538461539,
+      "wer": 0.25925925925925924
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.08018154311649017,
+      "wer": 0.3316062176165803
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2ebb65fd7a6699620faef9ed7fd11bfb8c118e8c0015d79acd547e73e574975
+size 138350251

preprocessor_config.json

@@ -0,0 +1,10 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0,
+  "processor_class": "Wav2Vec2Processor",
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe6ec7b3270aa367491157344ac2dcc8b4d111b6e29777ad2dc0db90ee290b02
+size 1262054513

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "<pad>", "do_lower_case": false, "word_delimiter_token": "|", "replace_word_delimiter_char": " ", "special_tokens_map_file": null, "name_or_path": "/tmp/tmpscoyck9o", "tokenizer_class": "Wav2Vec2CTCTokenizer", "processor_class": "Wav2Vec2Processor"}

vocab.json

@@ -0,0 +1 @@
+{"<pad>": 0, "<s>": 1, "</s>": 2, "<unk>": 3, "|": 4, "'": 5, "-": 6, "a": 7, "b": 8, "c": 9, "d": 10, "e": 11, "f": 12, "g": 13, "h": 14, "i": 15, "j": 16, "k": 17, "l": 18, "m": 19, "n": 20, "o": 21, "p": 22, "q": 23, "r": 24, "s": 25, "t": 26, "u": 27, "v": 28, "w": 29, "x": 30, "y": 31, "z": 32, "ß": 33, "á": 34, "ä": 35, "ö": 36, "ü": 37}